Re: [Tipc-discussion] Re: hang while deleting ports

[Mark H. <ma...@os...>]

On Tue, 2004-05-11 at 16:22, Jon Maloy wrote:
> I am not sure. In your first analysis of the problem you 
> found that it was a dying process that was causing the
> processor to hang while trying to grab the node lock, which
> was then presumably held by the tipc_rev_msg interrupt,
> or a tasklet.
> That is the opposite scenario of what you describe here.

I see the hang both ways.

> 
> Also, this was no different in the old implementation:
> we shared the same lock between user level/tasklet code
> and interrupts without any problems.  If you are right,
> why don't we have deadlocks all the time then ?

That is a good point, but it does seem a possibility.

> 
> I think I will re-read Rusty Lynch "unreliable guide" 
> once more...

That is Rusty Russell.


-- 
Mark Haverkamp <ma...@os...>